Establishment And Analysis Of Large-scale Image Quality Assessment Database

In people's daily life and learning,we often intentionally or unintentionally receive and perceive a variety of image information.Images are not only simply used for viewing,but also can convey a variety of information in more cases.Especially in the field of multimedia,images are given more meaning and play a greater role.In the field of image,image quality assessment plays an important role.No matter what image is processed,how it is transmitted or where it is applied,the word "quality" is necessary.The quality of an image is more like an attribute of an image,as people would say that the pixels of an image are too poor or the color is too saturated,and so on.People will also say that the quality of an image is too good or unsatisfactory.Therefore,image quality assessment as an important research field has been paid attention to.As a complete academic system,image quality assessment includes subjective and objective image quality assessment.These two distinct quality assessment methods evaluate images from different standpoints.Seemingly opposite research methods,in fact,they are mutually dependent and constrained.Only when subjective image quality assessment and objective image quality assessment are eventually returned to consistency,can the whole image quality assessment system be complete.In the field of image quality assessment,the starting point of research often comes from the published image quality assessment database.Image quality assessment database provides a broader stage and more abundant materials for the whole field of image quality assessment.Scientists use the open database to verify their own image quality assessment algorithm,and constantly improve the depth and breadth of image quality assessment.This paper will focus on the establishment and analysis of large-scale image quality assessment database.This paper divides the establishment of large-scale image quality assessment database into three stages: early stage,middle stage and late stage.Firstly,this paper discusses in detail the specific preparation for the early stage of database establishment,puts forward the requirements for higher updates of database,and tailors these requirements to design specific matters in the early stage.It includes the establishment and screening of original image,the selection of distortion type and the setting of distortion level.Whether the preparation is reasonable or not determines the quality and application scope of the whole image quality assessment database.The whole process is irreversible,and the database created in the earlier period will not be modified or processed in the middle and later period of establishment.Secondly,the experiment of subjective image quality assessment will be carried out on the large-scale image quality assessment database.The whole experiment process is very time-consuming and laborious.This paper will discuss in detail the test environment,test mechanism and test scheme of the experiment to restore the most real experimental process as far as possible.Hope to provide a better reference for the establishment of image quality assessment database in the future.Finally,the data recorded in the experiment will be sorted out and analyzed.Any excellent image quality assessment database is not created by randomly finding some images to generate different distorted images,or by randomly finding several people to evaluate the subjective quality of the database.On the basis of the former two,we select the mainstream objective image quality assessment indicators to evaluate the objective image quality.Through the consistency of the subjective and objective assessment results,we further illustrate that the large-scale image quality assessment database created in this paper is of great value and significance.